Local LLM Security

Binding a local model server to 0.0.0.0 hands the LAN an unauthenticated API

Ollama, LM Studio, vLLM and Gradio all listen on well-known ports, and none of them ask for a credential unless you go and build one, so the one-line change that lets you test from your phone also answers the office subnet.